During the third quarter of 1991 I spent most of my time on the reduction of binary star data from the MK3 interferometer on Mt Wilson. I started compiling all data files containing binary observations and, in the course of this, reduced about 100 more nights so that all observations which have been made up to date are calibrated (several nights need recalibration, however). Thus, we are now able to address the question of how many binaries that we have observed so far show a clear indication of their nature. 255 stars were originally selected from the catalogue of spectroscopic binaries by Batten, Fletcher and McCarthy. Out of 130 stars with a significant amount of data, 43 show visibility variations larger than about 20%. Orbits for five of them had been determined previously. Scheduling of new observations during the next months will concentrate on the remaining stars to find and improve their orbits (see below). Lower priority will be given to stars for which little data exists, in order to use the available time for observations in the most efficient way. During the second half of July I also was involved in tests of two seeing monitors, which were shipped in August to the MIRA observatory sites in California, near Monterey. The seeing monitors are modified portable 18 inch Newtonian telescopes which, by using a mask of two 2- inch apertures, measure the degree of turbulence in the atmosphere. Tests concerned the ability of the equatorial mounts to track stars and the quality of the output data. The tests were carried out at the USNO. At the beginning of August, I traveled to Mt Wilson for my scheduled term of observations. In addition, during that time I accompanied Dave Mozurkewich to MIRA (Monterey Institute for Research in Astronomy) in order to receive the two seeing monitors and to teach the staff of MIRA how to make the measurements that were planned for the following months. Apart from problems with the tracking ability at high wind speeds, the monitors functioned satisfactory. The observations on Mt Wilson were mostly succesful. Back in Washington, I started to work on the determination of orbits for the binaries which we had most data for. This has been done 'traditionally' in two steps: firstly, by determining the nightly mean values of the separation and of the position angle, and, secondly, by fitting an orbit to the (r,pa) values obtained over a period of time. It soon turned out that in order to improve the orbit and determining better values for the magnitude difference of the stars and their diameters, a program which fits all the relevant parameters directly to the visibilities was needed. In this way, data from all nights could be used (not only the ones which had sufficient data to allow for the determination of (r,pa) values), and orbits of short period binaries could be determined despite the considerable change in separation and in position angle during the night. In the first half of September I therefore developed a program to improve orbits using the described method and was able to determine five more orbits of binary stars. I also worked on a poster presentation of this work at the conference for high resolution imaging by interferometry at Munich, October 14 - 18.